The Definitive HDFS Service Start/Stop Checklist for Cloudera Clusters

[#1]

In a typical HDFS HA deployment, Zookeeper, ZKFailoverController, DataNodes, NameNodes and JournalNodes, are all installed. What's the best order of starting/stoping these HDFS services to avoid running into services dependency issues?

[#2]

Stop order:

1. Stop Datanodes

2. Stop Namenodes

3. Stop ZKs

4. Stop Journal nodes

Start Order:

1. Start ZKs

2. Start JNs

3. Start Datanodes

4. Start NNs

[#3]

If you are using ambari , It will take care of the sequence. You just need to click on start service and start service.

[#4 · Accepted solution]

You need to follow the below Start order

• Ranger
• Knox
• ZooKeeper
• HDFS
  • Journal Node
  • Name Node
  • Secondary Name Node
  • Data Node
• YARN
  • Resource Manager
  • History Server
  • Node Manager

[#5]

Thank you all for the answers.